Update to Epson scanning article: how to bypass all tonal adjustments [Archive]

Ken Lee

6-Jan-2018, 09:35

Ken,

Is that scanning a negative?, my version of Epson scan, doesn't actually work like that for negative. My version is the latest that for the 4990, so a bit old...

You can do the same in vuescan just by choosing the following:

173373

Keep in mind this will have sRGB gamma applied.

or you can just choose raw in viewscan which will give you a gamma of 1.

In the article, in the section (http://www.kennethleegallery.com/html/scanning/index.php#vssettings) where I deal with VueScan settings, I show that the default settings result in some clipping: I recommend other settings to circumvent the problem.

I am using an Epson V700 and it's been a while since I've updated my copy of VueScan.

Some people prefer to get nothing more than a "raw" scan from the scanner and perform all adjustments in Photoshop, Lightroom, etc. To date, no one has ever shared with me a method to get that unaltered scan using VueScan. I added this new section to the article after a reader contacted me with information about the MacOS Image Capture application. I did a little digging and found the corresponding Windows feature.

Ted Baker

6-Jan-2018, 09:43

In the article, in the section (http://www.kennethleegallery.com/html/scanning/index.php#vssettings) where I deal with VueScan settings, I show that the default settings result in some clipping: I recommend other settings to circumvent the problem.

I am pretty sure none of those setting in the current version would do what you want.

The setting I gave should result in no additional changes, other than sRGB gamma being applied, but I don't bother in any case and just scan it as raw file.

When you have it right the graph should look like this:

173374

To get a raw scan in vuescan just click, the raw scan option:

173379

This will NOT have any gamma corrections applied. Vuescan also gives the option to adjust the CCD exposure time manually.

Anyway I hope that helps :)

Ken Lee

6-Jan-2018, 10:54

The VueScan raw file is an output file type, like JPG, PNG, etc. It is a TIFF embedded in a file, with some additional header information, if I understand correctly. In that sense it is like the raw files we get from digital cameras: no compression.

Vuescan automatically adjusts image brightness as we crop. We can't bypass that. Note that the settings on the left-hand side remain unchanged as we crop.

http://www.kennethleegallery.com/images/forum/VueScan1.png
Here our crop includes a lot of white background: VueScan has darkened the scan to accommodate the high average tone.

http://www.kennethleegallery.com/images/forum/VueScan2.png
Here our crop includes mostly middle grays: VueScan has lightened the scan accordingly.

With the Epson drivers, cropping the image also creates tonal adjustments... but we can dispose of them if we like. We can't do that with VueScan.
http://www.kennethleegallery.com/images/forum/Epson1.png

http://www.kennethleegallery.com/images/forum/Epson2.png

With the generic drivers provided by Microsoft and Apple, no matter how we crop the image, the resulting scan is the same. Adjustments are not applied.

Ted Baker

6-Jan-2018, 11:07

The VueScan raw file is an output file type, like JPG, PNG, etc. It is a TIFF embedded in a file, with some additional header information, if I understand correctly. In that sense it is like the raw files we get from digital cameras: no compression.

You don't understand correctly. But I can't say I blame you... Vuescan is not the most accessible program in any case..

I hope you don't take that the wrong way. Your site is very inspirational!

IanBarber

6-Jan-2018, 15:11

You might find this update helpful: How to Bypass All Tonal Adjustments (http://www.kennethleegallery.com/html/scanning/index.php#bypass).

It shows how to get an image directly from an Epson scanner with no tonal adjustments, using the MacOS and Windows generic drivers as well as Epson Scan software.

If anyone knows how to do this with VueScan (or would like to share any other helpful suggestions or corrections), please let me know so that the article will be more complete and informative.

Thanks !

How does the Epson Scan deal with Gamma with these settings. Does it still encode in gamma 2.2 or is there a way to force it to give a true Linear scan with a Gamma of 1.0

To produce a Linear Scan with VueScan to be used with something like ColorPerfect (ColorNeg) we can use these settings

173396 173397 173398 173395

Leigh

6-Jan-2018, 15:18

I am using an Epson V700 and it's been a while since I've updated my copy of VueScan.
Some people prefer to get nothing more than a "raw" scan from the scanner and perform all adjustments in Photoshop, Lightroom, etc. To date, no one has ever shared with me a method to get that unaltered scan using VueScan.
Why not use EpsonScan rather than VueScan.

EpsonScan came with my V750. I'm sure it also came with your V700.

For a plain-vanilla scan, you don't need any of the fancy VueScan features, so why use it?

- Leigh

IanBarber

6-Jan-2018, 15:21

Vuescan automatically adjusts image brightness as we crop. We can't bypass that. Note that the settings on the left-hand side remain unchanged as we crop.

Ken, In VueScan, doesn't Crop Buffer % value determine how VueScan adjusts the brightness

Leigh

6-Jan-2018, 15:25

The VueScan raw file is an output file type, like JPG, PNG, etc.
It is a TIFF embedded in a file, with some additional header information, if I understand correctly.
Sorry, Ken,

But that's not possible.

JPG et al are compression formats.

Even if you choose "lossless compression", you still get pixel mangling.

TIFF is a direct element-to-pixel dump of the sensor's raw data. If you do a TIFF from a scanner or sensor, the file size is the same as the sensor pixel count plus minor overhead.

- Leigh

Ken Lee

6-Jan-2018, 15:50

Ken, In VueScan, doesn't Crop Buffer % value determine how VueScan adjusts the brightness

Even if you set it to 0, the brightness of the image changes as you vary the cropping.

Ken Lee

6-Jan-2018, 15:53

Why not use EpsonScan rather than VueScan.

EpsonScan came with my V750. I'm sure it also came with your V700.

For a plain-vanilla scan, you don't need any of the fancy VueScan features, so why use it?

- Leigh

Yes, that's one of the points made in the article: I don't use VueScan. It's a rather long article, so that may have escaped people's attention if they only had time to skim it, or skim this thread for that matter.

I have simply added some information to the article, for those people who have asked me how to make a plain scan with no adjustments whatsoever.

IanBarber

6-Jan-2018, 16:03

Even if you set it to 0, the brightness of the image changes as you vary the cropping.

I will do some testing with this tomorrow especially with the crop buffer and report back my findings

Ted Baker

6-Jan-2018, 16:16

How does the Epson Scan deal with Gamma with these settings. Does it still encode in gamma 2.2 or is there a way to force it to give a true Linear scan with a Gamma of 1.0

When you scan in epson scan with the gamma set to 1.0, the file you output will be encoded into the sRGB gamma (most likely 100% correctly) unless you say differently I think you can also use adobeRGB which has similar gamma. You can convert it easily enough to 1.0, a couple of different ways, and you should get exactly the same thing, before epson scan applies the correction. ColorPerfect I understand offers this feature. You only need a gamma of 1.0 if you intend to do your own image processing, in which case you can convert it yourself as long as you know the original gamma. When you use tools like LR they will automatically convert it a gamma of 1.0 internally to store in its own working space.

Ted Baker

6-Jan-2018, 16:23

Vuescan automatically adjusts image brightness as we crop.

What your probably seeing there is vuescan simulating the exposure change, that it intends to make when you actually do the scan. Vuescan actually allows you manual control this, and I think it also allows you do this for each color channel. It certainly allows the adjustment of the later. Just like changing the shutter speed, changing the exposure time of the CCD is not a tonal change.

There is no manual control of this in epson scan, though presumably it makes use of this feature, most probably it has calculated it before you are presented with the preview.

IanBarber

6-Jan-2018, 16:38

When you scan in epson scan with the gamma set to 1.0, the file you output will be encoded into the sRGB gamma (most likely 100% correctly) unless you say differently I think you can also use adobeRGB which has similar gamma.

Ted, are you saying that even if you choose gamma 1.0 in Epson Scan, it will converting along the way to gamma 2.2 which is what (sRGB or AdobeRGB 1998) are

Ted Baker

6-Jan-2018, 16:48

Ted, are you saying that even if you choose gamma 1.0 in Epson Scan, it will converting along the way to gamma 2.2 which is what (sRGB or AdobeRGB 1998) are

Yes, otherwise you would not be a happy photographer :) BTW sRGB is not exactly 2.2 but close enough, same for adobeRGB.

In addition I assume we are talking about scanning positives, because scanning a negative is different.

IanBarber

7-Jan-2018, 03:21

Yes, otherwise you would not be a happy photographer :) BTW sRGB is not exactly 2.2 but close enough, same for adobeRGB.

In addition I assume we are talking about scanning positives, because scanning a negative is different.

I was referring to scanning black and white negatives

Ted Baker

7-Jan-2018, 04:00

I was referring to scanning black and white negatives

In that case the gamma etc, still applies. But the method of inversion is potentially unique to each program, it may not be a big secret, the point being no vendor that I am aware of publishes it. For example what is the method used by epson? How this is done most certainly effects the tonal relationship, this is most obvious in colour photography, but it also applies in BW.

If you do a raw scan (Vuescan/Silverfast/DSLR raw file), a negative will come out looking like a negative, with no gamma applied. If you use epson scan you can scan a negative as a positive, and remove the sRGB gamma later. I believe colorperfect supports the later.

If you stay in 16bit integer, you can potentially unravel any tonal changes, exactly if you know what the original changes were (assuming you round up/down correctly). Clipping is the exception of course, but that is not a tonal change. If you use a RGB curve tool in software later down the line you can also unravel those changes, (or indeed make your own), and get back to the original potentially. I hope this makes some sense.

Ken Lee

7-Jan-2018, 06:55

What your probably seeing there is vuescan simulating the exposure change, that it intends to make when you actually do the scan. Vuescan actually allows you manual control this, and I think it also allows you do this for each color channel. It certainly allows the adjustment of the later. Just like changing the shutter speed, changing the exposure time of the CCD is not a tonal change.

There is no manual control of this in Epson scan, though presumably it makes use of this feature, most probably it has calculated it before you are presented with the preview.

I think I finally grasp what you are saying. Duhhh :rolleyes:

I made a scan with VueScan and configured it to output both a raw file and a tiff file. The tonal scale of the tiff file was affected by cropping and adjustments made in the Color tab like curve low, curve high. In addition, a gamma adjustment of approximately 2.2 has been applied to the tiff file. The resulting tiff file matches what we see in the preview pane.

The raw file, on the other hand, looks rather dark, because not even a gamma correction has been applied. It's just the original scanner acquisition of the image.

http://www.kennethleegallery.com/images/forum/vuescantiffandraw1.png

If we apply a gamma correction to the raw file, it begins to resemble the TIFF image (which may also contain tweaks from the Color tab and cropping).

http://www.kennethleegallery.com/images/forum/vuescantiffandraw2.png

Finally, the Epson scan, Apple Image Capture and Windows Scan drivers are all silently applying a gamma correction to the scan. Turning off all corrections in the Epson drivers, we get an image which matches Apple Image Capture and Windows Scan, but none of these 3 approaches actually delivers the original scanner capture because a gamma adjustment has been applied in all 3 cases. In fact, the only configuration which delivers the original scanner capture, is the VueScan raw file.

...Is this correct ?

IanBarber

7-Jan-2018, 07:41

Ken, after you scanned the image as a RAW file in VueScan and brought it into Photoshop, what option did you choose on the profile Mismatch dialog box ?

Ken Lee

7-Jan-2018, 07:53

Ken, after you scanned the image as a RAW file in VueScan and brought it into Photoshop, what option did you choose on the profile Mismatch dialog box ?

I don't get a profile mismatch prompt because I have them all turned off.

I'm no expert but my workflow is rather simple, 99% monochrome except for snapshots.

For printing I work in Gray Gamma 2.2 because that's what Piezography uses with Quadtone RIP. For making JPG files for the web I use sRGB.

http://www.kennethleegallery.com/images/forum/ColorSettings.png

Tin Can

7-Jan-2018, 08:12

Good work guys.

IanBarber

7-Jan-2018, 08:20

My guess is that when you bring the RAW scan into Photoshop, it will be an untagged negative. To get it to a positive, are you just inverting the image and then applying the Gamma 2.2

Ken Lee

7-Jan-2018, 08:31

My guess is that when you bring the RAW scan into Photoshop, it will be an untagged negative. To get it to a positive, are you just inverting the image and then applying the Gamma 2.2

In these tests I have made low-res scans of a positive (a recent print) because it's easy to spot a faithful rendition when we can see the original. It's easy to tell that the 3 scans made with no adjustments (Epson with adjustments disabled, Apple Image Capture and Windows Scan) all resemble the original print and closely match one another. Similarly, it's easy to see that the VueScan raw file is dramatically different, as no Gamma adjustment has been applied.

Normally after scanning a b&w negative I explicitly convert it to Gray Gamma 2.2 and save it in Photoshop PSD format. (When acquiring a raw file from my digital camera, I do the same thing.) I perform all corrections and printing in Gray Gamma 2.2.

If I make a JPG to put on a web page, I convert to sRGB.

chassis

7-Jan-2018, 12:34

Ted Baker

7-Jan-2018, 12:57

I think I finally grasp what you are saying. Duhhh :rolleyes:

...

Finally, the Epson scan, Apple Image Capture and Windows Scan drivers are all silently applying a gamma correction to the scan. Turning off all corrections in the Epson drivers, we get an image which matches Apple Image Capture and Windows Scan, but none of these 3 approaches actually delivers the original scanner capture because a gamma adjustment has been applied in all 3 cases. In fact, the only configuration which delivers the original scanner capture, is the VueScan raw file.

...Is this correct ?

Yes of those options the Vuescan raw, is the only one unchanged in anyway. (There are a few ways to save raw in vuescan, that will allow minor corrections BTW).

I don't know anything about Apple Image Capture and Windows Scan, as they are 8 Bit?, I won't comment on them as they are probably a distraction.

Back to epsonscan, it is my understanding that if you scan a positive per you instructions, it will be same as the viewscan raw with the following adjustments:

1. the sRGB gamma applied
2. the colors are adjusted to match the sRGB color coordinates.

The later is point is very minor, I asked Ed Hamrick (vuescan author) about it, but he is not keen on divulging those kind of details. To be fair I can't imagine epson would help either...

But it is important to note, that if you are dealing with a negative, all the scanning programs will apply their respective algorithms to invert the negative.

Most people want an image in sRGB, or something very similar in anycase, so it's important to understand the context.

Ken Lee

7-Jan-2018, 13:19

Ken, this is tremendous, thanks for updating your tutorial and sharing it. For black and white, I had migrated to the 0-255 slider settings, influenced by my C-41 work.

For gamma, with C-41, do you have suggestions? With color negative, correcting for the orange mask means either adding or subtracting data (or multiplying or dividing, depending on your workflow). In this situation, clipping, particularly of highlights, needs to be avoided. It is challenging.

Have you done tests with Epson Scan gamma set at values other than 1.0 or 2.2?

I have scanned color negatives at gamma 3.0 and 4.0, all other settings per your tutorial, save as .tiff. Higher gamma in Epson Scan shifts the negative histogram to the right. For box-speed-exposed Portra 400, the histogram falls almost in the middle, which is helpful for subsequent color correction work. I find that box speed exposure of Portra 400 with gamma 1.0 or 2.2 easily results in clipped highlights after color correction (orange mask and color balance).

Does Epson Scan gamma setting change data acquisition from the sensor, or is it a software manipulation?

Thanks again, this is valuable and helpful.

You're welcome and I very much appreciate the helpful information shared by others here and elsewhere. Ted has provided several valuable missing links.

I've done very little scanning of color materials: hopefully there are tutorials on the web which cover the issues. Positives are hard enough, but negatives are more challenging because we never know what they "really" look like unless we scan a calibration target (hint hint).

Being merely a consumer of these consumer-grade Epson scanners, it's not clear to me whether any adjustments we make via scanner drivers affect the process before image acquisition (aperture, exposure time, lamp brightness in physical terms or gain or other attributes in electronic terms) or whether all the adjustments we request are performed in software, after image acquisition. To be certain, you'd have to know the API (application program interface), namely the protocol by which the driver software talks to the scanner. Making matters worse, not all APIs are published by the manufacturer: sometimes vendors publish only the basics and keep the good ones for their own use, or make them available only by license, to protect their intellectual property.

IanBarber

7-Jan-2018, 13:22

From my experience with VueScan when scanning black and white negatives.

When setting the Output to Raw even if you make any additional changes to exposure, curve adjustments etc, those adjustments are not carried over when you press scan.

The scanned image (negative) should have a gamma encoding of 1.0. When importing to Photoshop, it should be assigned the same gamma value of your grey working space (gray gamma 2.2)

Now, the image needs to be inverted from a negative to a positive and this is where things in my opinion can get tricky. A simple invert does not seem to do a very good job but the ColorPerfect plugin on the other hand does a very good job with the inversion.

Andrew O'Neill

8-Jan-2018, 12:16

Jeez all the years I've been scanning, I've been scanning RAW... but no clue what gray gamma PS is set at. I'll have to check when I get home.

faberryman

8-Jan-2018, 14:02

I believe the default is Gray Gamma 2.2, but you might want to double-check your settings.

8-Jan-2018, 17:12

Alan Klein

8-Jan-2018, 18:54

You're welcome and I very much appreciate the helpful information shared by others here and elsewhere. Ted has provided several valuable missing links.

I've done very little scanning of color materials: hopefully there are tutorials on the web which cover the issues. Positives are hard enough, but negatives are more challenging because we never know what they "really" look like unless we scan a calibration target (hint hint).

Being merely a consumer of these consumer-grade Epson scanners, it's not clear to me whether any adjustments we make via scanner drivers affect the process before image acquisition (aperture, exposure time, lamp brightness in physical terms or gain or other attributes in electronic terms) or whether all the adjustments we request are performed in software, after image acquisition. To be certain, you'd have to know the API (application program interface), namely the protocol by which the driver software talks to the scanner. Making matters worse, not all APIs are published by the manufacturer: sometimes vendors publish only the basics and keep the good ones for their own use, or make them available only by license, to protect their intellectual property.

Ken: I use a V600 with Epson Scan software. I have never used other scanner software.

My take is that all changes are done after the scan. Basically we're dealing with post scan processing that could be done with Photoshop or Lightroom or whatever. The point is the hardware is designed for the highest dMax. Assuming the program can change the amplification of the sensors or the amount of light output, it would have to be less than the standard operating scanner settings set to maximum or best. That would make shadows area darker. No point in doing that.

Increasing the light, even if possible, doesn't make sense either. You have to assume Epson designed and constructed the machine to give us the best dMax with the sensors and light they furnished. Why would they give us less? So you can't really raise the light output or increase the sensor amps without distortion.

The other giveaway to me that adjustments are all in post is when you do a test scan. You can play with the settings and get examples of what you will get once you do the actual full scan. The only way the program can give you those examples is by processing the test quick scan in software. So it seems, they do exactly the same thing after the full scan - software processing. Frankly, the argument other scanner programs are doing something more than Epson during the scan is sales promotion. If you like scan programs that you can "post" adjust when scanning, fine. But you can't get blood from a turnip. Why learn a second post processing program after learning PS or LR?

The only question about a settings that might be useful is by setting the black and white points (levels) to where the actual data starts and ends. That may give you more data of the parts of the film that actually contain something. But I'm not sure of that. It might make no difference by leaving levels at 0 and 255. When I try it both ways, it appears to be no difference. But some people have claimed there is a difference. I;m not sure about this. Maybe someone has an explanation of what we get with level settings.

Alan Klein

8-Jan-2018, 19:07

By test quick scan in my previous post, I meant preview scan.

Alan Klein

8-Jan-2018, 19:17

I'm not sure if it's different because I have a PC, but putting the sliders to the ends makes the preview go black. So flat, it's not linear, but flat black. I can sorta get what you have shown by moving the white slider a little left to the start of the histogram highlights. Just a minor gotcha, but I know what you're doing and do something similar to your normal procedure but adjust gamma in PS. I do each color separately as my negatives are pyro stained and blue/red get shifted a little.

173505173506

First, the negative appears very dark, hence it's black with 0-255. My scans of good negatives are usually more to the left side. However, if you try moving the left slider in post just to the left of the start of data, you should find the results has the same brightness and contrast as if you did it with the Epson Scan during the scan process. The reason is like I said in my last two posts. The Epson software is just applying Levels to the resultant scanned file just like you would with a post processing program applied to a flat scan.

faberryman

8-Jan-2018, 19:29

There seems to be much conflicting information.

Ken Lee

8-Jan-2018, 19:55

Ken: I use a V600 with Epson Scan software. I have never used other scanner software.

My take is that all changes are done after the scan. Basically we're dealing with post scan processing that could be done with Photoshop or Lightroom or whatever. The point is the hardware is designed for the highest dMax. Assuming the program can change the amplification of the sensors or the amount of light output, it would have to be less than the standard operating scanner settings set to maximum or best. That would make shadows area darker. No point in doing that.

Increasing the light, even if possible, doesn't make sense either. You have to assume Epson designed and constructed the machine to give us the best dMax with the sensors and light they furnished. Why would they give us less? So you can't really raise the light output or increase the sensor amps without distortion.

The other giveaway to me that adjustments are all in post is when you do a test scan. You can play with the settings and get examples of what you will get once you do the actual full scan. The only way the program can give you those examples is by processing the test quick scan in software. So it seems, they do exactly the same thing after the full scan - software processing. Frankly, the argument other scanner programs are doing something more than Epson during the scan is sales promotion. If you like scan programs that you can "post" adjust when scanning, fine. But you can't get blood from a turnip. Why learn a second post processing program after learning PS or LR?

The only question about a settings that might be useful is by setting the black and white points (levels) to where the actual data starts and ends. That may give you more data of the parts of the film that actually contain something. But I'm not sure of that. It might make no difference by leaving levels at 0 and 255. When I try it both ways, it appears to be no difference. But some people have claimed there is a difference. I;m not sure about this. Maybe someone has an explanation of what we get with level settings.

Excellent points and well stated... thank you !

Ted Baker

8-Jan-2018, 21:46

it's not clear to me whether any adjustments we make via scanner drivers affect the process before image acquisition (aperture, exposure time, lamp brightness in physical terms or gain or other attributes in electronic terms) or whether all the adjustments we request are performed in software, after image acquisition.

There are no direct controls available that affect the BEFORE image capture in Epsonscan unlike vuescan. In vuescan you can manual adjust the CCD exposure, time. You even measure it with your stop watch, if you use a gain of 2, it will take twice as long to scan the image, as the stepper motor slows down to accommodate the longer time. However like the moon landing, it's possible this is a fake but I doubt that it is ;) Like the shutter speed on you camera if you set it too long, you will blow the highlights.

As I said earlier, epson scan most likely sets this automatically in the preview. Manual control of exposure allows you to skip the preview stage, in vuescan.

You could actually test this in epson scan if you wanted too by scanning film that has a very dense base, and timing it. (There is another clever trick you can try too, put a matching film clip over the calibration area).

There is also the individual gain of the each channel, vuescan also allows you to control manually, my understanding is that this is analogue amplification, like changing the iso on your digital camera.

All of this is hidden in vuescan, unless you actually look for it, as it will do this function automatically, as I believe epsonscan does anyway.

I would suggest most probably aren't interested, if the automatic function works correctly, it's already complicated enough :)

From my experience with VueScan when scanning black and white negatives.

When setting the Output to Raw even if you make any additional changes to exposure, curve adjustments etc, those adjustments are not carried over when you press scan.

That's the idea with raw, i.e. to remove all the digital transformations. Even if setup it up to do the inversion, the preview may be positive but the raw file will still be a negative. However the gain functions I mentioned above will affect the raw capture. There is also some tweaks to the raw mode, that will allow you some minor processing.

Some of the features in viewscan are woefull, but for certain things it very simple and efficient (if what it offers, matches what you want and you know how to use it).

All of his doesn't mean you can't achieve excellent/similar results with epsonscan, especially if you are doing ALL of your post processing later.

IanBarber

9-Jan-2018, 06:01

That's the idea with raw, i.e. to remove all the digital transformations. Even if setup it up to do the inversion, the preview may be positive but the raw file will still be a negative. However the gain functions I mentioned above will affect the raw capture. There is also some tweaks to the raw mode, that will allow you some minor processing.

Does the Gain feature work with Epson scanners, I thought this was a feature of the Nikon scanners same as the Focus feature

Ken Lee

9-Jan-2018, 08:33

In vuescan you can manual adjust the CCD exposure, time. You even measure it with your stop watch, if you use a gain of 2, it will take twice as long to scan the image, as the stepper motor slows down to accommodate the longer time.

Could you please share where this is set ?

There is also the individual gain of the each channel, vuescan also allows you to control manually, my understanding is that this is analogue amplification, like changing the iso on your digital camera.

All of this is hidden in vuescan, unless you actually look for it

I see where we can affect the color balance of the 3 channels...

http://www.kennethleegallery.com/images/forum/vuescanred.jpg

...but the raw file appears to be unaffected by changes to these settings. If both images are captured at the same time, during a single pass, could gain/exposure time be changed for one but not the other ? Isn't this color gain adjustment performed post-capture ?

I can't get any changes to color balance to result in a different raw file: whatever I do, the raw file looks the same. What have I overlooked please ? Thank you for your patience.

http://www.kennethleegallery.com/images/forum/viewscanredandraw.jpg

http://www.kennethleegallery.com/images/forum/vuescanblueandraw.jpg

IanBarber

9-Jan-2018, 09:13

I think Ted is referring to the Gain control on the Input Tab

173512

Ken Lee

9-Jan-2018, 09:38

I think Ted is referring to the Gain control on the Input Tab

173512

Eureka !

Those options appear if we scan transparencies. And as promised, the scanner stepper motor runs slower when we set them to a higher value.

Eureka !

And as promised, the raw file becomes darker or lighter accordingly.

Thank you again for your patience.

IanBarber

9-Jan-2018, 09:58

Sometimes if I need to open up the shadows slightly, I will increase the brightness of the Green channel found under the Color Tab

This is with the Green brightness set to 1 (default)

https://www.ianbarberphotography.co.uk/lff/green-1.jpg

This is with the Green brightness set to 2

https://www.ianbarberphotography.co.uk/lff/green-2.jpg

Think of the red Green Blue brightness controls as gamma adjustment (middle slider in Photoshop levels)

https://www.ianbarberphotography.co.uk/lff/green-brightness.jpg

chassis

9-Jan-2018, 10:11

Good discsussion folks. Ian or Ted, are you using color negative film with any of your scanning work? The ideas on setting gain or gamma for each channel are interesting to me. It seems consistent with the different speeds/responses of the several color emulsion layers in C-41 film. I can see this difference in the histograms in Epson Scan and in post.

Ken Lee

9-Jan-2018, 10:24

I'm not sure if it's different because I have a PC, but putting the sliders to the ends makes the preview go black. So flat, it's not linear, but flat black. I can sorta get what you have shown by moving the white slider a little left to the start of the histogram highlights. Just a minor gotcha, but I know what you're doing and do something similar to your normal procedure but adjust gamma in PS. I do each color separately as my negatives are pyro stained and blue/red get shifted a little.

173505173506

This isn't because you're using a PC, it's because you're scanning a transparency. If I scan a transparency on my Mac I get the same black image once I move the right slider too far to the right.

In my previous tests I scanned a reflective image: apparently the interface gives different results depending on what we're scanning. Arghhh !

Making matters worse, the 3 tools (Histogram Adjustment, Tone Correction and Image Adjustment) appear to have a cumulative effect. Using one doesn't override the other.

Therefore, when scanning transparencies with Epson Scan software, I can't find a way to bypass corrections at all.

As Gomer Pyle used to say, "Surprise, surprise" !

Ted Baker

9-Jan-2018, 10:42

Eureka !

Those options appear if we scan transparencies. And as promised, the scanner stepper motor runs slower when we set them to a higher value.

Eureka !

And as promised, the raw file becomes darker or lighter accordingly.

Thank you again for your patience.

Yes, thems the ones. A hint is they are all under the input tab ;)

I can't do a screen print as I need to be connected to a scanner.

These are slope adjustments (in linear colorspace), i.e. 2X will have twice the changes to the highlights (numerically), as opposed to the shadows, just like changing the ISO on your camera. (2x 50% becomes 100%, 18% becomes 38%, 5% becomes %10 etc)

With the color gain, it's possible this is being faked (done after the ADC), but the scanner has this capability, it does this every time automatically as the ccd goes over the calibration area, put a colour film clip over this area and see how it affects the results.

I don't know what the stuff on color tab actually does, other than to set it to leave everything alone, other than the sRGB adjustment. (But it's ignored completely when raw scanning which is what I do). All the stuff under the color tab is post scan processing that can be done somewhere else.

The ideas on setting gain or gamma for each channel are interesting to me.

Be a little carefull here, lest you be come totally confused.;)

This isn't because you're using a PC, it's because you're scanning a transparency. If I scan a transparency on my Mac I get the same black image once I move the right slider too far to the right.

...

Therefore, when scanning transparencies with Epson Scan software, I can't find a way to bypass corrections at all.

As mentioned previously an inversion is by definition a tonal change. How this works is potentially unique to each program. This is very obvious when you try and work in color. Each program works their magic or NOT :D

IanBarber

9-Jan-2018, 11:50

Alan Klein

9-Jan-2018, 22:56

Ian, I copied your first darker picture and applied lightening to the shadows in PS Elements and got basically the second picture. So basically, it's just a post processing edit that the scanner is doing, not magic during the scan.

173540

Alan Klein

9-Jan-2018, 23:05

This isn't because you're using a PC, it's because you're scanning a transparency. If I scan a transparency on my Mac I get the same black image once I move the right slider too far to the right.

In my previous tests I scanned a reflective image: apparently the interface gives different results depending on what we're scanning. Arghhh !

Making matters worse, the 3 tools (Histogram Adjustment, Tone Correction and Image Adjustment) appear to have a cumulative effect. Using one doesn't override the other.

Therefore, when scanning transparencies with Epson Scan software, I can't find a way to bypass corrections at all.

As Gomer Pyle used to say, "Surprise, surprise" !

If you want to scan without any adjustments with Epson Scan, uncheck all boxes. Then click on the Configuration button at the bottom. When it switches to the Configuration page click on the No Color Corrections. The scan. Then do all your changes on the resultant "flat" file with your post processing program.

IanBarber

10-Jan-2018, 04:09

Ken Lee

10-Jan-2018, 05:51

If you want to scan without any adjustments with Epson Scan, uncheck all boxes. Then click on the Configuration button at the bottom. When it switches to the Configuration page click on the No Color Corrections. The scan. Then do all your changes on the resultant "flat" file with your post processing program.

Eureka ! That works nicely.

Given what Ted has pointed out, I guess it shouldn't be too surprising that the VueScan raw file and the Epson Scan file made with no color corrections don't match exactly.

Scanning a B&W negative, the Epson version comes out as a positive where some kind of gamma adjustment has already been applied. The VueScan raw version comes out as a negative and requires a Gamma 2.2 adjustment followed by an inversion to positive. The Epson version is "snappier", while the VueScan version is flatter. Which version is a more faithful rendition of the negative, who can say ?

When scanning a reflective image with Epson Scan, there's a discernible difference between a scan made with "No Color Corrections" and one made where we zero-out the Histogram Adjustment settings... Oh well :rolleyes:

Perhaps this thread should be renamed to "how to minimize tonal adjustments".

Alan Klein

10-Jan-2018, 09:27

Eureka ! That works nicely.

Given what Ted has pointed out, I guess it shouldn't be too surprising that the VueScan raw file and the Epson Scan file made with no color corrections don't match exactly.

Scanning a B&W negative, the Epson version comes out as a positive where some kind of gamma adjustment has already been applied. The VueScan raw version comes out as a negative and requires a Gamma 2.2 adjustment followed by an inversion to positive. The Epson version is "snappier", while the VueScan version is flatter. Which version is a more faithful rendition of the negative, who can say ?

When scanning a reflective image with Epson Scan, there's a discernible difference between a scan made with "No Color Corrections" and one made where we zero-out the Histogram Adjustment settings... Oh well :rolleyes:

Perhaps this thread should be renamed to "how to minimize tonal adjustments".

I believe you're right that you still have control of the Levels (black and white points) even when No Color Correction is checked off. Going back to the COnfiguration page, there are setting for controlling the gamma and other stuff other than No Color Correction or color correction. I don;t know anything about how they work though.

Ken Lee

10-Jan-2018, 09:43

I believe you're right that you still have control of the Levels (black and white points) even when No Color Correction is checked off. Going back to the COnfiguration page, there are setting for controlling the gamma and other stuff other than No Color Correction or color correction. I don;t know anything about how they work though.

http://www.kennethleegallery.com/images/forum/EpsonSlider.jpg

You're right that the slider can still be adjusted, but I just tested both a reflective and negative scan: no matter how we set the Auto Exposure Level slider, the resulting images are indistinguishable. Once we select No Color Correction, it doesn't matter what else we do: everything else is disabled and greyed-out. (I suspect the fact that the Auto Exposure Level slider still slides, is an oversight: the software developer forgot to disable the slider.)

Ted Baker

10-Jan-2018, 10:58

no matter how we set the Auto Exposure Level slider, the resulting images are indistinguishable.

I think you misunderstand the purpose of exposure in this context. If I was to look at a series of photographic prints, printed traditionally, with different camera exposures how would I tell the difference? How would a picture taken at f16 and 1/500 on hp5 look different from f8 at 1/500?

Ken Lee

10-Jan-2018, 11:10

Ted Baker

10-Jan-2018, 11:19

We would see a corresponding difference in the density of the resulting images.

In the prints? Expand the example take a series of shots from -5 stops underexposed through to +10, print say a mid tone in the picture at the same density. Would the prints all look the same? would some look very similar to the "correct" exposure?

Pere Casals

10-Jan-2018, 11:47

I think you misunderstand the purpose of exposure in this context. If I was to look at a series of photographic prints, printed traditionally, with different camera exposures how would I tell the difference? How would a picture taken at f16 and 1/500 on hp5 look different from f8 at 1/500?

But if exposure was changed then scanning speed would also change, so it would be noticed...

IMHO there is a way to check that the scanner is actually bypassing all tonal adjustments, this is scanning an Stouffer wedge (T2115 or T4110). Then knowing the calibrated densities we can calculate (MS Excel) the theoric progresion related to lowest density, or from a mid grey, the calculated theoric values have to match a linear scan reading for each wedge step, every 2 wedge steps (or 3 in the T4110 case) the reading should be the half or the double, the reading can be seen with Ps, Information panel shows the value for the spot we point with the mouse...

Exposure should be adjusted to a level that it does not saturate the CCD in an empty area, but this perhaps is not enough to get an accurate reading for deep shadows in slide film, no problem with negative film (BW or color) , but slides may have very dense areas that (if important) may require Multi-Exposure (single pass or multiple pass) to get an accurate reading, still the calculated reading should be linear.

Ted Baker

10-Jan-2018, 11:51

But if exposure was changed then scanning speed would also change, so it would be noticed...

Which is what we established vuescan offers and you can measure it, epsonscan does NOT have a manual method to vary the exposure...

IMHO there is a way to check that the scanner is actually bypassing all tonal adjustments, this is scanning an Stouffer wedge (T2115 or T4110). Then knowing the calibrated densities we can calculate (MS Excel) the theoric progresion related to lowest density, or from a mid grey, the calculated theoric values have to match a linear scan reading for each wedge step, every 2 wedge steps (or 3 in the T4110 case) the reading should be the half or the double, the reading can be seen with Ps, Information panel shows the value for the spot we point with the mouse...

That will work for positive scan, it won't work for a negative or at least it shouldn't because you DO want a tonal adjustment.

Ken Lee

10-Jan-2018, 11:56

In the prints? Expand the example take a series of shots from -5 stops underexposed through to +10, print say a mid tone in the picture at the same density. Would the prints all look the same? would some look very similar to the "correct" exposure?

The prints would not look all the same. Some would be similar to the correct exposure.

Pere Casals

10-Jan-2018, 12:34

That will work for positive scan, it won't work for a negative or at least it shouldn't because you DO want a tonal adjustment.

For WB in reality I dont wan't tonal adjusment for the negatives, in any case I prefer to shape the curve with Ps. I always scan with V850 in that way, taking all linear range .

For WB I'm working with the possibility to take a linear scan and then using PS to preview the effect of selective color masking http://phototechmag.com/selective-masking-part-iii-computer-techniques-for-the-traditional-darkroom/

The idea is painting in a layer with contrast filter colors over the linear image. Then a LUT should convert from the coloured image to a soft-proofing image that would predict the result if printing the coloured mask and using it sandwiched with the negative in the enlarger.

The challenge is the calibration software that takes sensitometric tests and generates the LUT. I clarify, the LUT converts the linear image painted over with magenta to yellow colors to a greyscale showing the reflective density predicted for each spot.

This is to design a color mask WYSWYG to get local contrast control in the darkroom, and to print that mask with a cheap inkjet. The sensitometric response is to be measured with printed color samples, made with the same inkjet and using same colors from PS...

Ted Baker

10-Jan-2018, 12:35

The prints would not look all the same. Some would be similar to the correct exposure.

Exactly and you are an experienced photographer, so you can imagine why manual control is not offered in epson scan, and indeed why an automatic function can do a very good job of the required task. It's not clear in epson scan what high or low even means, or how it varies the automatic exposure calculation, if at all. Indeed it may just be a mistake.

Ted Baker

10-Jan-2018, 12:46

For WB in reality I dont wan't tonal adjusment for the negatives, in any case I prefer to shape the curve with Ps. I always scan with V850 in that way, taking all linear range .

I am sure that's what everyone wants, but that is not what happens. If I print a Black and white print on grade 2 paper, I WILL get a tonal adjustment from the negative, and that is what I want and that IS THE WAY it is designed. (any grade actually).

If I want to process a roll of negative film as a positive, other than the reversal step (bleach/2nd developer), what else do you need to do? ;)

I hope you understand I am not criticising your methods in anyway, I am just point out a discrepancy in your description.

Ari

10-Jan-2018, 14:30

bob carnie

10-Jan-2018, 14:39

That was always the goal with Epson scanners, Alan; get a neutral scan, flat as can be, so that one could have more post-processing options available in PS/LR.
I've since moved to an Eversmart Pro, but the scanning goal is the same: neutral, flat, unsharpened file that can be tweaked as much as I like in post.

I have the Eversmart Supreme as well and I agree with Ari, 16 bit also Gamma 2.2 for BW 16bit RGB (I use Adobe 1998) and keep within the scanners native resolution, rather than trying to make bigger files.

Tin Can

10-Jan-2018, 14:50

A D750 and many other cameras have a setting for Flat, Neutral, Standard, Vivid, Monochrome, Portrait and Landscape.

The video guys want 'Flat' for the best in post.

I do like this thread. :)

10-Jan-2018, 15:14

Ken, thank you for your webpage explaining scanning! It's good and this thread only helps everyone figure things out also.

Pere Casals

10-Jan-2018, 18:28

I am sure that's what everyone wants, but that is not what happens. If I print a Black and white print on grade 2 paper, I WILL get a tonal adjustment from the negative, and that is what I want and that IS THE WAY it is designed. (any grade actually).

If I want to process a roll of negative film as a positive, other than the reversal step (bleach/2nd developer), what else do you need to do? ;)

I hope you understand I am not criticising your methods in anyway, I am just point out a discrepancy in your description.

There is no discrepancy...

The real scene or what is recorded in the negative can have more range than what the display medium allows, a paper has 1:100 aprox static contrast. So a common situation is having to compress (or cut off) more or less the shadows and the highlights to have mids taking enough dynamic range. This can be performed by the scanner software automaticly or it can be adjusted manually with Ps. For important shots I prefer to adjust manually what's in the toe and what's in the shoulder, tonal adjustment with PS is flexible and easy, single key point is having a 16 bits tiff file to avoid banding.

I also guess this is a very common practice for film photographers: scanning as raw as possible (a flat scan, with no sharpen) and cooking with edition software.

Ted Baker

11-Jan-2018, 04:25

There is no discrepancy...

You never answered the question. :)

What you describe IS a tonal adjustment.

Pere Casals

11-Jan-2018, 06:12

What you describe IS a tonal adjustment.

Yes, of course, but one thing if making tonal adjustment with scanner software and another one is doing it in Ps. I was speaking about that, as my understanding is that the thread is about that: "how to bypass all tonal adjustments" is to do it with edition.

Tonal adjustments in scanner software may work great a lot of times, but if wanting full control best is doing that in edition soft...

IanBarber

11-Jan-2018, 06:29

Although it hasn't been mentioned from what I can see, has anyone actually done a scan with no adjustments of a stouffer wedge to see how many steps the scanner can actually read

Ken Lee

11-Jan-2018, 07:19

I can do this later today (in my time zone).

IanBarber

11-Jan-2018, 07:21

I can do this later today (in my time zone).

Thanks Ken, looks forward to the results

Pere Casals

11-Jan-2018, 08:01

Although it hasn't been mentioned from what I can see, has anyone actually done a scan with no adjustments of a stouffer wedge to see how many steps the scanner can actually read

Here there is an scan of a Kodak Q14 target: https://www.flickr.com/photos/timparkin/26276532831/in/faves-125592977@N05/ , including grey scale.

This is a test made by photographer Tim Parkin. You can download a 15000 pix wide original: https://www.flickr.com/photos/timparkin/26276532831/sizes/l

If scanning the stouffer with the Epson take care to mask the empty area to avoid any stray light, then you can make the same scan again without masking to see the impact of stray light. I would place the wedge in the long scanning direction to avoid most stray light, in this way higher densities are illuminated alone (when emply area masked).

The Tim Parkin scan shows how powerful is Velvia and how difficult is to display that greatness with present electronic displays, and shows the need of "multiexposure" feature in scanners.

IMHO Parkin is not only a great photographer, but also a master in Velvia science !

Ken Lee

11-Jan-2018, 08:14

Thanks Ken, looks forward to the results

Back in 2013 (maybe earlier) I reported on this test but did not use the "flat" or "raw" methods discussed in this thread.

See http://www.largeformatphotography.info/forum/showthread.php?101499-Ideal-density-for-scanning

IanBarber

11-Jan-2018, 08:16

Pere Casals

11-Jan-2018, 08:31

Back in 2013 (maybe earlier) I reported on this test but did not use the "flat" or "raw" methods discussed in this thread.

See http://www.largeformatphotography.info/forum/showthread.php?101499-Ideal-density-for-scanning

Ken, this is an interesting test, and anyway 3.0D is a lot for negative film.

...but I would ask if Multiexposure feature was used in that test. In my experience ME (with Silverfast) boosts max density. Also it is important how dense is the rest of the row that is scanned at the same time, if all the row is dense then we have less stray light messing in very dense spots.

My experience is consistent with what LS says: http://www.silverfast.com/highlights/multi-exposure/en.html

http://www.silverfast.com/img/highlights/multiexposure/SilverFast_ME_big_en.jpg

When scanning 135 velvia film with V850 I also noticed a difference if scanning a single strip and masking the other 3 slots in the holder. In extreme situations IMHO V850/750 have some (less stray light) advantage over 800/700 because 850/750 lenses are coated.

For most SF versions for V750/700 one has to purchase that feature separately.

Well, drums have that advantage, as a single point is iluminated then stray light dispears, the superior DR of PMT can be overcomed with ME, but stray light from the rest of the scanned row cannot be avoided.

I was thinking that perhaps flatbeds could be improved (since now have LED illuminators) with the choice of illuminating and reading smaller sections of the row, at the cost of slower scanning.

Ted Baker

11-Jan-2018, 12:16

Back in 2013 (maybe earlier) I reported on this test but did not use the "flat" or "raw" methods discussed in this thread.

See http://www.largeformatphotography.info/forum/showthread.php?101499-Ideal-density-for-scanning

If you want to do this properly, to see what the scanner is capable of, you need to adjust the exposure time, so that film base, of the wedge is very close to "white" or 2^16. If you want do the same for color neg, you also adjust the color gain to make the base "white" as well.

You could potentially go back then and use epsonscan, and see if you can achieve the same results, using the algorithm used by it's auto exposure.

I did this a while back with a fuji step wedge,but I can't find the file at the moment.

Two things worth remembering you should measure the results digitally, you want be able to see the differences with your eye, on a monitor. The second is that the requirements for a negative are different. It is very possible automatic setting are close to optimal in this regard.

My experience is consistent with what LS says: http://www.silverfast.com/highlights/multi-exposure/en.html

That, link also mentions the controversial exposure time adjustment, that epson scanners DO HAVE... :D

Pere Casals

11-Jan-2018, 13:33

That, link also mentions the controversial exposure time adjustment, that epson scanners DO HAVE... :D

IMHO there are patents involved. https://www.google.com/patents/US7102679 , this one is about noise , but perhaps other is about DR.

An scanner without "exposure time adjustment" it can also benefit from multiple exposures as averaging readings lowers noise.

IMHO a good technique is making multiple exposures with different exposure times with a single pass, this is making more than one exposure before the scanner advances to the next row. It looks that the Epsons can only make multiple exposures if making multiple passes, each pass with a different exposure time... Is this because patents ?

More expensive scanners can make multiple exposures in a single pass.

Multiple passes can be good in some cases because the super-resolution effect...

Ted Baker

11-Jan-2018, 14:06

IMHO a good technique is making multiple exposures with different exposure times with a single pass, this is making more than one exposure before the scanner advances to the next row. It looks that the Epsons can only make multiple exposures if making multiple passes, each pass with a different exposure time... Is this because patents ?

Who knows for you sure, but these devices have not been advanced in a long time, where as you probably have a device in your pocket that is quite quite capable of doing what is described in that patent.

IMHO it's important not to bundle all your requirements for scanning under one banner as slides and negatives have different requirements.

Also this link has the DR results for v700 and 4990

http://www.silverfast.com/PDF/TestReport_ME_DWueller.pdf

It puts it as 3.1 for a single pass. That is quite a lot when you adjust the time, to zero out the film base.

Ken Lee

11-Jan-2018, 14:33

Here's a quick test of EpsonScan with no adjustments and VueScan raw with Gamma 2.2 added. There are links to TIFF files below if you're interested. Settings will be shown in another post since there's a limit to the number of images per post.

For the VueScan file I added a Gamma 2.2 layer adjustment in Photoshop. I flipped both images for easy reading. Otherwise the TIFF files are direct from the scanner. The JPGs are for quick reference. Your perception of these JPG files will depend on your monitor and its calibration (http://www.kennethleegallery.com/html/tech/index.php#calibrate).

I don't have time to perform any analysis or plotting of the values: perhaps someone would like to do that :cool:

Epson Scan with no adjustments JPG:

http://www.kennethleegallery.com/images/forum/EpsonScanNoAdjustments.jpg

VueScan raw file with Gamma 2.2 adjustment JPG:

http://www.kennethleegallery.com/images/forum/VueScanRawWithGamma.jpg

Epson Scan with no adjustments TIFF here (http://www.kennethleegallery.com/images/forum/EpsonScanNoAdjustments1200.tif)

VueScan raw file with Gamma 2.2 adjustment TIFF here (http://www.kennethleegallery.com/images/forum/VueScanRawWithGamma1200.tif)

VueScan raw file TIFF here (http://www.kennethleegallery.com/images/forum/VueScanOriginal1200.tif)

Ken Lee

11-Jan-2018, 14:33

Epson Scan settings:

http://www.kennethleegallery.com/images/forum/EpsonStepWedgeSettings.png

VueScan settings:

http://www.kennethleegallery.com/images/forum/VueScanStepWedgeSettings.png

Pere Casals

11-Jan-2018, 14:45

Also this link has the DR results for v700 and 4990

http://www.silverfast.com/PDF/TestReport_ME_DWueller.pdf

It puts it as 3.1 for a single pass. That is quite a lot when you adjust the time, to zero out the film base.

And 3.4D with multiexposure it is a pretty competent job !

Ted Baker

11-Jan-2018, 14:49

Quick review,

Vuescan exposure is much better, very close to perfect (what is the density of number 1 supposed to be?). From the example I assume you just used auto and slide film, vuescan does a good job in that mode.

Epson scan not so good, but it is also possible to make mistakes in the analysis, especially with Epsonscan can you post the original vuescan file, without any changes? Working from that you should be able to work out what epsonscan does.

When you did your cropping in preview for both programs, did you make sure that there was no clearer areas than step 1?

Then of course you need to it for all three colors with a filter if you want to check that aspect of it as well.:D

Ken Lee

11-Jan-2018, 14:57

Quick review,

Vuescan exposure is much better, very close to perfect (what is the density of number 1 supposed to be?). From the example I assume you just used auto and slide film, vuescan does a good job in that mode.

Epson scan not so good, but it is also possible to make mistakes in the analysis, especially with Epsonscan can you post the original vuescan file, without any changes? Working from that you should be able to work out what epsonscan does.

Then of course you need to it for all three colors with a filter if you want to check that aspect of it as well.:D

I have added a link the original raw VueScan scan file with no gamma adjustment to the same post where I posted the others.

All those files are down-sized to a width of 1200 pixels.

IanBarber

11-Jan-2018, 15:00

Ken, just remind me why you chose Slide Film as the Media in Vuescan

Ted Baker

11-Jan-2018, 15:04

Thanks,

Will take a look and do a proper analysis over the next few, days. (my own step wedge was old and only went to 2.1)

If you have a chance to do it again, with vuescan do the preview, put a crop around step 3 or step 6 (roughly the density of BW film base). Hit preview again, the click on lock exposure, then crop the entire step wedge. (note the exposure time for reference)

Steps below your reference will be blown out, but it will be interesting to check for any improvement in SnR in darker steps.

Actually this last test is quite important, as it should show if there is any meaningful difference for BW negative scanning.

Pere Casals

11-Jan-2018, 15:05

Here's a quick test of EpsonScan with no adjustments

Nice test !

By adjusting the curve it is made clear that the Epson reaches well 3.05D, that this wedge reaches, but with some noise in the higher, IMHO if using multiexposure noise would impreve a lot and a denser wedge would be needed to see the scanner limit.

By inspecting the values (F8, information dialog, 16 bits value) the expected value for step 4 should be 1/2 if the step 1, it goes close but with some discrepance... perhaps actual densities shpuld be checked or taken from a calibration...

173598

Ken Lee

11-Jan-2018, 15:11

Ken, just remind me why you chose Slide Film as the Media in Vuescan

I guessed that scanning it as a slide, IE a positive, would be simpler and therefore better, since no reversal step is included in the process.

Ted Baker

11-Jan-2018, 15:18

I guessed that scanning it as a slide, IE a positive, would be simpler and therefore better, since no reversal step is included in the process.

Ken, you made the best choice for the test. When using auto exposure, which is the default, unless you go looking for manual exposure. the auto exposure algorithm is different. When you choose negative it will adjust all three channels, but this is "ISO" change not a time change and actual exposure time it calculates is different. (Its shorter, probably because shorter is faster). With Raw you still get a negative regardless.

chassis

11-Jan-2018, 15:36

Ken, thanks for this, it is a big service to the community here.

Ken Lee

11-Jan-2018, 15:43

For the record this was a Stouffer Transmission Step Wedge sized 4x5 inches. According to the packaging it has 21 steps in "increments of .15".

If my math is correct, 21 steps amounts to a density range of 10.5 f/stops. 2 exp(10.5) = 1448, which in log (10) terms is 3.16.

Ted Baker

11-Jan-2018, 17:24

Most importantly if you look at those samples, yourself you see a reduction in noise, quite significant actually.

A simple method to view this is:

use a curve tool, to invert and brighten the darker steps. (adjust the intensity so yo can see the noise, I did about 75% on LAB scale)

173601

If ken has time, and we are very lucky... he might post an example with vuescan, using an ideal typical exposure for BW negative stock. (using the method suggested earlier)

But before you rush out to rescan all your negs;), remember this will only affect highlight or brighter mid tones in a dense negative. And a lot this probably disappears under the "weight" of 4x5, but very relevant to 35mm.

Ken Lee

11-Jan-2018, 18:28

If you have a chance to do it again, with vuescan do the preview, put a crop around step 3 or step 6 (roughly the density of BW film base). Hit preview again, the click on lock exposure, then crop the entire step wedge. (note the exposure time for reference)

Steps below your reference will be blown out, but it will be interesting to check for any improvement in SnR in darker steps.

Actually this last test is quite important, as it should show if there is any meaningful difference for BW negative scanning.

VueScan based on Level 3 with Gamma 2.2 correction JPG:

http://www.kennethleegallery.com/images/forum/VueScanBasedOnLevel3WithGamma.jpg

Settings:

http://www.kennethleegallery.com/images/forum/VueScanBasedOnLevel3Settings.jpg

VueScan based on Level 3 with Gamma 2.2 correction TIFF here (http://www.kennethleegallery.com/images/forum/VueScanBasedOnLevel3Settings.jpg)

VueScan based on Level 3 no correction TIFF here (http://www.kennethleegallery.com/images/forum/VueScanBasedOnLevel3.tif)

Ken Lee

11-Jan-2018, 18:49

before you rush out to rescan all your negs;), remember this will only affect highlight or brighter mid tones in a dense negative. And a lot this probably disappears under the "weight" of 4x5, but very relevant to 35mm.

After applying a Gamma 2.2 correction to the VueScan raw file (not adjusted to level 3, the original scan), it seems that the VueScan values are somewhat more equi-distant than the Epson values. I guess this means the VueScan scan is more linear. But what if we apply a slight correction curve to the Epson scan: will any differences become negligible ?

What does it suggest that the lowest VueScan values are shifted to the right ?

http://www.kennethleegallery.com/images/forum/VueScanCurve.jpg http://www.kennethleegallery.com/images/forum/EpsonCurve.jpg

Ted Baker

11-Jan-2018, 22:58

After applying a Gamma 2.2 correction to the VueScan raw file (not adjusted to level 3, the original scan), it seems that the VueScan values are somewhat more equi-distant than the Epson values. I guess this means the VueScan scan is more linear. But what if we apply a slight correction curve to the Epson scan: will any differences become negligible ?

You need to careful that you don't jump to the wrong conclusion. You have applied a gamma to these, so they will not be linear, and they are not. It is also probable that the gamma 2.2 function you applied does not match sRGB gamma function applied in the epson scan. What this tests shows is what the dynamic range of the scanner is, and how to make a small, and possibly imperceivable improvement.

Interesting it matches with the silverfast paper.

What does it suggest that the lowest VueScan values are shifted to the right ?

That's to be expected as you have made 2.53 X (1.2 stops) exposure increase, but this is not a tonal change as all the tonal ratio's remain the same. What you are doing is keeping the highlights away from the noise floor.

This is screen shot of the last three steps at 85%, which is a pretty extreme example, but I can see the difference in noise.

173604

173605

Pere Casals

12-Jan-2018, 02:33

That's to be expected as you have made 2.3 stops exposure increase,

Why is it 2.3 stops ? steps are 1/3...

IanBarber

12-Jan-2018, 03:00

Ted Baker

12-Jan-2018, 03:54

Why is it 2.3 stops ? steps are 1/3...

Careless on my part :o, Ken set the exposure time to 2.53 times increase, which I read as 2.3 the first time. I used stops out of habit.

Which I guess underscores the need to review the analysis.

2.53 X is a lot by the way, you won't be able to go that high normally. For 35mm fomapan 400 I use 2. The film has a fairly dense base. For hp5 on 120 1.2x is the max you can use. For 4x5 I measure it.

Ted Baker

12-Jan-2018, 04:04

Should both not be in the same colourspace, either sRGB or GrayGamma 2.2 to see a better evaluation?. My understanding is that sRGB and GrayGamma are 2 different beasts.

Its helpfull that they are in same colorspace. I converted vuescan to sRGB , and epson one to a linear scan, but could not find an easy way to post it here. Will try a bit harder tommorow.

Pere Casals

12-Jan-2018, 04:04

Careless on my part :o, Ken set the exposure time to 2.53 times increase, which I read as 2.3 the first time. I used stops out of habit.

Which I guess underscores the need to review the analysis.

2.53 X is a lot by the way, you won't be able to go that high normally. For 35mm fomapan 400 I use 2. The film has fairly dense base. For hp5 on 120 1.2x is the max you can use. For 4x5 I measure it.

Easy to fix, just replace the . by a / :), 2.3 to 2/3

foen

12-Jan-2018, 08:06

I don't understand yet if could be a routine workflow for scans or not

Pere Casals

12-Jan-2018, 09:30

I don't understand yet if could be a routine workflow for scans or not

It depends on if you want a routine scanning workflow or not, or better said when you want it or not.

IMHO when you want to make an extensive Photoshop edition/optimization for the image best is making a routine scanning workflow, taking all, and adjusting tonality and etc in Ps.

Also IMHO if one wants to just do some clicks in Lightroom then we can adjust the image with scanner software because we don't plan complex adjustments than may need all information in the negative.

Ken Lee

12-Jan-2018, 11:26

The Stouffer Transmission Step Wedge has 21 steps in "increments of .15". Each step is a half f/stop.

If my math is correct, 21 steps amounts to a density range of 10.5 f/stops. 2 exp(10.5) = 1448, which in log (10) terms is 3.16. If a scanner can clearly separate all steps on this target, it can handle a dMax of 3.16

When I tested the Epson scanner using Epson software a number of years back (http://www.largeformatphotography.info/forum/showthread.php?101499-Ideal-density-for-scanning) with my usual method of adjusting the histogram, I found that the scanner could separate "16 steps, or 8 stops, which is 256 levels, or in log 10 terms, 2.4".

Can either of the "raw" scans I made yesterday properly separate all 21 steps, or for that matter, significantly more than 16 steps ? By properly, I mean in a linear manner where each step is equidistant from the next.

Pere Casals

12-Jan-2018, 11:41

Yes you are right, the 21 model is 1/2 stops steps, the 31 model is 1/3...

...but 21 steps are 20 difference, this is 10 stops, so 21 step is 1/(2^10) the 1st step, this is 1/1024. A range of near exactly 3.0D difference, the stouffer info says 3.05D (http://www.stouffer.net/Photo.htm), but the .05 may be because base density of the wedge.

Ted Baker

12-Jan-2018, 14:16

I don't understand yet if could be a routine workflow for scans or not

This is just test, the negative has yet to be inverted. When it is, a tonal adjustment will occur. This adjustment will be dependant on the method used by software that performs this change.

Can either of the "raw" scans I made yesterday properly separate all 21 steps, or for that matter, significantly more than 16 steps ? By properly, I mean in a linear manner where each step is equidistant from the next.

The scanner can do at least 3.05D maybe it can even do 4D like the marketing suggests... :eek: But the tests seems to show how the noise seems to build up, and how a small adjustment to exposure can make a difference.

the stouffer info says 3.05D (http://www.stouffer.net/Photo.htm), but the .05 may be because base density of the wedge.

That may be true, ken made a 2.53X adjustment, which is probably about 1.3 stops (if the vuescan units are meaningful), "wiping" out the first three steps on the wedge.

Ken Lee

13-Jan-2018, 18:03

Funny - My Epson V700 just died. I hadn't used it much lately until making these tests, but it now puts out a funny noise and the red lights flash.

No good deed goes unpunished I guess :rolleyes:

I took my old 4990 of the basement shelf and fired it up. Works fine. I wonder if it will cost more to repair the 700 than the price of a new unit.

Pere Casals

13-Jan-2018, 18:38

Funny - My Epson V700 just died. I hadn't used it much lately until making these tests, but it now puts out a funny noise and the red lights flash.

.

there are 2 sliders to block the scanner for transport, one in the illuminator and another for the bed, check
both are in the right position

Ken Lee

13-Jan-2018, 18:48

Thanks Pere. I checked and reset those switches several times.

I was making test scans all day: at a certain point the machine just stopped working.

I presume that some part needs replacement.

Tin Can

13-Jan-2018, 18:52

Ken, I bet you got plenty of mileage from that pile of plastic. You shoot way more than I. I bought a V700 and not the higher models on your website recommendation and tutorials. I have been very happy with mine. It even has survived two moves. That surprised me.

The only good thing about a new one is that the LED does not need a warmup. Or maybe that's a bad thing on second thought. Warmup is good for many things.

I reserve my V700 for only LF negs. Prints I use a scanner app on iPhone.

IanBarber

14-Jan-2018, 02:46

This is just test, the negative has yet to be inverted. When it is, a tonal adjustment will occur. This adjustment will be dependant on the method used by software that performs this change.

Do we know what method the ColorNeg/ColorPerfect plugin uses as this to me seems to do a far better job at the inversion than Photoshop, just a shame about the clunky interface

IanBarber

14-Jan-2018, 02:49

Funny - My Epson V700 just died. I hadn't used it much lately until making these tests, but it now puts out a funny noise and the red lights flash.

No good deed goes unpunished I guess :rolleyes:

This happened to my V800 3 weeks ago. According to Epson, the error was because it could not communicate with the drivers. I deleted the drivers (MAC) reinstalled them and its been ok since, still cannot work out if/why drivers would cause the error.

Ted Baker

14-Jan-2018, 03:06

Do we know what method the ColorNeg/ColorPerfect plugin uses as this to me seems to do a far better job at the inversion than Photoshop, just a shame about the clunky interface

The company published some of the maths behind their product. I used it as starting point for my own project, along with some stuff from Arri and Kodak. I think the Photoshop inversion is just (1-X) where x is normalised value of the negative scan, that is just a guess, but if it is poor then that is probably why...

I guess this means the VueScan scan is more linear. But what if we apply a slight correction curve to the Epson scan: will any differences become negligible ?

I went through each of the scans that ken posted, I first converted them to a gamma of 1, and measured each step. I can't seem to post the spreadsheet, they all show a reasonably linear response from the CCD, it is less linear towards the bottom of the scale. I think the vuescan exposure is clearly better, it gets you further away from the noise floor, and the response is more linear away from the bottom of the scale. Can you tell the difference in a real print? I don't know! I suspect maybe with colour, where poor accuracy in the highlights would result in colour casts.

Pere Casals

14-Jan-2018, 03:21

I checked and reset those switches several times.

the drivers

Yes Ian, you are right, first is downloading last drivers and reinstalling.

Anyway an abnormal noise may happen when stepper motors cannot move the carriage because it is blocked, perhaps if the scanner was operated while it was blocked it could happen than a wheel jumped a tooth or something like that, resulting in a missalignment. I would remove the bed glass and I'd inspect what happens, then I'd free mechanical links to see if a section of the mechanical chain of the drive is mechanically blocked, or perhaps some dry teflon lube can solve it. Because the noise, it may be a mechanical issue easier to DIY fix.

The only good thing about a new one is that the LED does not need a warmup.

Randy, you are right, V850/800 have same performance, avoiding the warmup is convenient, but IMHO there is another advantage, LED illuminators are very stable requiring little or no calibration overtime, if one wants the scanner calibrated. Lamps of the V700/750 (I guess) can change illumination a bit at the beiginning, then there is a more stable level during several thousands of hours, and by end of life of lamps again we have changes in the illuination's level and hue.

My V750 came with an IT8 calibration slide, the V850 came without it. Perhaps with saving the cost of the IT8 target they can fit the LEDs :), well, and sure today leds are cheaper...

Ken Lee

14-Jan-2018, 06:45

The scanner makes the unpleasant sound as soon as the power switch is turned on. It's not a software issue as far as I know.

I guess I need to open it up and see if there's anything obviously wrong. I'm not particularly good at repairs.

chassis

14-Jan-2018, 06:46

Ken, sorry to hear about the V700. Not sure if this helpful, but I think you are on a mac platform. I recently installed High Sierra and it caused a problem communicating with a 4 year old Epson all-in-one printer/scanner. Normal wireless installation of the printer is no longer possible, and it needed to be installed using a fixed IP address. I now use this all-in-one device wirelessly using the fixed IP address. EpsonScan no longer functions using this connection method, and Apple ImageCapture is the only way to utilize the scanner for simple/everyday things.

I use my V750 with a USB cable. While on the surface this may seem apples and oranges, as mentioned above, the drivers and connectivity could be the issue. I would also add that any OS updates done recently can be problematic, if you are using mac. The recent update to High Sierra was also somewhat painful to install, after a very painful High Sierra installation.

Ken Lee

14-Jan-2018, 06:51

I don't think it's not a software problem. The scanner worked fine while I was doing tests. Then it just stopped working after no changes were made to anything. It can only be a mechanical or electronic failure.

Pere Casals

14-Jan-2018, 08:06

It can only be a mechanical or electronic failure.

Just a thought...

There are two "zero position" sensors for homing, one is for the sensor carriage and the other is for the illuminator carriage, perhaps when "powering on" the scanner reaches the mechanical top without sensing the zero position, and thus trying to go beyond, so the stepper motor would make that noise because it could not rotate.

First try to see if the noise comes from the under bed drive or from the illuminator drive in the cover. Then locate the "end of travel" sensors that should tell the control electronics that a carriage its at home. The scanner uses those sensors to find a known starting position, from there it can count steps to place the sensor and illuminator in the wanted position. I don't know what kind of sensor is it, but for sure you have 2 there...

There several videos in youtube explaining V700/750 disassembly for cleaning https://www.youtube.com/watch?v=z7QkfM5TQ4A

Anyway if you find that a carriage tries to go beyond a mechanical end it would be worth to repair it, because probably parts to replace are simple.